Co-Simulation testbed and controllers for compressor-based loads providing grid balancing services

Description

This code has been developed for the ARPA-E project "Overcoming the Technical Challenges of Coordinating Distributed Load Resources at Scale". Its purpose is to simulate a feeder with residential A/Cs that are being controlled for the provision of load balancing services. The simulation environment can run stand-alone, where voltages are set artificially, or it can run linked with GridLAB-D, where GridLAB-D simulates the distribution network down to the house nodes, and where the MATLAB code simulates the Thermostatically Controlled Loads (TCLs) that are connected to the distribution feeder.

How to run

NOTE: If cloning the repository locally in a Windows machine fails due to the length of some file names, you can run Git Bash as an administrator and execute the command: git config --global core.longpaths true

You will first need to download the repo from https://github.com/gridlab-d/TaxonomyFeeders and unzip it into a folder called "TaxonomyFeeders-master" located in the main directory of this codebase (the directory that includes the main_runSim script).

The following three scripts need to be executed sequentially. 1. pre1constructFeeders: constructs a user-defined set of PNNL feeders from feeder files located in TaxonomyFeeders. 2. pre2tclSetup: constructs the house models. 3. mainrunSim: runs a simulation with the specified controller. * To simulate the distribution feeder, useGLD should be set to 1. In that case, the script will generate a batch file that can be executed from the command line and will run a simulation using Matlab and GridLAB-D. * Otherwise, useGLD should be set to 0. In that case, the simulation can be run directly within Matlab as no link with GridLAB-D will be used.

Main Options

• Simulation timestep: Determines how often the temperature dynamics and on/off modes are updated. Can be specified in pre1_constructFeeders through the timeStepInSec variable.
• Population size: Determines the size of the controlled population. Can be specified implicitely in pre1_constructFeeders through the feederOptions.percentTclOptionSet variable.
• Mixture of 2-zone houses: Determines what percentage of the controlled population corresponds to houses with 2 zones and 1 comporessor. Can be specified in pre1constructFeeders through the perc2zone variable.
• Feeder head voltage: Determines the pu voltage value at the feeder head. Can be specified pre1_constructFeeders though the feederOptions.voltageRegulatorOptionSet variable.
• Feeder file: Determines what feeder to use. Can be specified pre1constructFeeders though the feederOptions.setOfTaxFiles variable. Should match the corresponding choices in pre2tclSetup and main_runSim.
• Scenarios for controlled houses: Determines the ambient conditions, building parameter distribution and heterogeneity, lockout duration, setpoint ranges. All these can be specified in pre2tclSetup and should match the choices in mainrunSim.
• Distribution network option: Option to run with or without a feeder model. If network voltages are required, GridLAB-D will be used. This option can be specified in main_runSim through the simulationOptions.useGLD variable.
• Simulation duration: Duration in hours for the simulation. Can be specified in main_runSim through the simulationOptions.simLength variable.
• Communication network: Option to consider delays and packet dropouts in the communication network. Can be controlled by changing the simulationOptionSet.commScenario variable in main_runSim.
• Controller timestep: Determines at what resolution the controllers will be acting. Can be specified in main_runSim through the simulationOptions.controlTimeStep variable.
• Controller type: There are three main controllers that can be currently used, a Markov-based controller, a Packetized Energy Management (PEM)-based controller and a baseline PID controller. Can be specified in main_runSim through the simulationOptions.controllerType variable.
• Controller parameters: The specific variables associated with each controller can be set in controllerSetup.
• Tracking signal options: The are 4 types of regulation signals that can be used to assess controller performance, including PJM RegD signals. The type and amplitude of the tracking signals can be set in main_runSim.

Authors

• G. S. Ledva, University of Michigan (now at Virtual Peaker, Inc.)
• I. M. Granitsas, University of Michigan
• O. Oyefeso, University of Michigan
• S. A. Nugroho, University of Michigan (now at Cummins, Inc.)
• P. Phanivong, University of California at Berkeley
• I. A. Hiskens, University of Michigan
• J. L. Mathieu , University of Michigan

License

This project is licensed under a GNU General Public License. Details are provided in license.txt.